Gallery — Project / Lab Report

This article proposes to obtain a statistical model of the daily peak electricity load of a household located in Austin-TX,USA. The Box-Jenkins methodology was followed to obtain the best fit for the time-series. Four models provided a good fit: ARIMA(0,1,2), ARIMA(1,1,2), SARIMA(0,1,2)(0,1,1) and SARIMA(1,1,2)(0,1,1). The model with the highest Akaike Information Criteria was the ARIMA(1,2,2). However, the model with the highest forecast accuracy was the SARIMA(1,1,2)(0,1,1), which obtained an RMSE of 0.296 and a MAPE Of 15.00.

Engineering notebook, outlining the design process of an autonomous robot competing in the 2016 TYESA/ASEE robot parade competition.

Report on laboratory practical

This project explores the predictability of play-calling of the Simpson College football offense during the years of 2012-2014. The models used to assess the predictability include simple analysis as well as linear and logistic regression.

Report without cover page Created by Zengjie Xia on 30/01/2018. Copyright © 2018 Zengjie Xia. All rights reserved.

Math project.

In most corruption scandals, the use of front companies for money laundering is almost ubiquitous. This work proposes to apply image classification to detect such organizations, through the use of Convolutional Neural Networks (CNN), namely the AlexNet architecture. The images are obtained by address search in Google Street View API, and the resulting classification will be further used along with other features to detect front com- panies in order to help the auditors from the Ministry of Transparency and Office of the Comptroller General (CGU, in Portuguese). To this moment, we applied classification to almost 15 thousand suppliers scenes with active contracts with the Brazilian Government until September 2016, obtained through data matching between the Government Purchases database and the Brazilian Federal Revenue Office database (more recent scenes should be added as this work progresses). Preliminary results with a pre-trained AlexNet CNN show the need for developing new scene classes more suited to the Brazilian context. In order to do this, we propose to apply clustering algorithms in features extracted from the last fully-connected layer of this net. The classes obtained will be used to fine-tune the AlexNet CNN for future classification, through the use of training from scratch or fine tuning techniques.

This is the boilerplate I used for my third co-op report at the University of Victoria.

Prática 9 de Tópicos de Física - CEFET